Rules VS Associations

Question 1

INTELLIGENCE

Answer 1

• much debate even w/humans
• one thing aka. general purpose ability (ie. Spearman’s g) VS social/verbal/spatial skill composite
• how to apply latter to animals?
• humans = IQ tests; animals = dif problems
• look for rule use ability evidence distinct from performance based w/associative learning

Question 2

SERIAL REVERSAL LEARNING

Answer 2

MACKINTOSH (1974)
- ability to improve over reversals = IQ index w/some studies
- problem learned to 90% criterion (correct colour occurs either side)
- reverses (colour matters not position) & repeat
RESULT
- later reversals = less error in acquiring discrimination (sometimes only 1)

Question 3

SERIAL REVERSAL LEARNING: ANALYSIS

Answer 3

• result = consistent w/rules in animals
• win-stay/lose-shift rule: pick colour -> reward -> keep picking same colour; no reward -> switch
• BUT other explanations must be rules out
• maybe animal = learning attendable stimulus aspects (ie. colour > position) -> enhances learning BUT single trial?
• transfer test required to counter objection

Question 4

LEARNING SETS

Answer 4

HARLOW (1949)
- improvement rate over problems = IQ index
- serial reversal repeat but w/dif shapes not colours
RESULT
- later problems = less errors in acquiring discrimination (sometimes only 1); further WS/LS rule evidence

Question 5

LEARNING SETS: ANALYSIS

Answer 5

• learning to attend certain stimulus features won’t explain results as stimuli/relevant features change between problem
• 1-trial acquisition suggests WS/LS rule
• BUT other possibilities ie. obtaining reinforcer = cue to solve discrimination (recent cue = good w/reinforcer BUT bad w/o aka. conditional discrimination)
• animals can learn dif colour stimuli = rewarded w/dif tones (ie. click = yellow; tone = blue)
• no urgent postulation need; what is dif between conditional discrimination VS learning sets?

Question 6

TRANSITIVE INFERENCE

Answer 6

• can perform transitive inference aka. if A > B > C, is A > C?
• can animals do this? can it be evidence for some relational rule use?
• ANS = transitive inference evidence in animals; some explained more simply BUT other = v persuasive
• test design trains on pairwise item chain -> novel pairing

Question 7

TRANSITIVE INTERFERENCE: MCGONIGLE & CHALMERS (1977, 1992)

Answer 7

• train = A+/B-; B+/C-; etc.
• monkeys trained on each stimuli pair; learn each one -> tested on novel pair (ie. B VS D)
• B chosen at STATSIG aka. learned A > B > C etc. aka. B > D
• transitive interference example

Question 8

TRANSITIVE INTERFERENCE: VON FERSEN ET AL. (1991)

Answer 8

• associative explanation
• A-E train ; A = always reinforced; E = never reinforced; B VS D test
• aka. generalisation from various pairs causes association; B/D both reinforced/non half time
• BUT B associated w/A (always reinforced)/ D associated w/E (never reinforced) -> B > D

Question 9

TRANSITIVE INTERFERENCE: TREICHLER & VAL TILBURG (1996)

Answer 9

• rhesus monkeys; attempt to rule out associative explanation; unconvincing
• train = A+/B-; B+/C-; etc.
• 2 problem sets trained up; linked w/E+/F- (train end) trials
• should prefer B > G; got this result
• BUT doesn’t work w/o linking training
• linking training lowers associative strength

Question 10

ANALOGICAL REASONING

Answer 10

• puppy -> dog = lamb -> ?
• can animals do this? YES = relational rule evidence ie. juvenile form (above)
• ANS = NO; bar…
  SARAH: PREMACK’S LANGUAGE CHIMP
• communicates w/symbols for “same/dif”

Question 11

ANALOGICAL REASONING: GILLAN ET AL (1981)

Answer 11

SARAH: PREMACK’S LANGUAGE CHIMP
- shown key/padlock; “same” symbol placed
- closed paint can placed; paintbrush/can opener choice
- picked can opener FIRST TIME aka. key/padlock = pain can/can opener
- did same w/pen/paper, etc.
- MacPhail = due to extensive language training over years

Question 12

SUMMARY

Answer 12

ANIMAL IQ…
- undoubtedly impressive abilities
- BUT many explained w/associative learning appeal or innate special skills (ie. homing pigeons)
- BUT still some evidence consistent w/rule use
- some rare animals can reason analogically (ie. Sarah)

Question 13

SUMMARY

Answer 13

ANIMAL IQ…
- undoubtedly impressive abilities
- BUT many explained w/associative learning appeal or innate special skills (ie. homing pigeons)
- BUT still some evidence consistent w/rule use
- some animals can reason analogically (ie. Sarah)